Monitoring experimental results

No matter whether an experiment is a success, a failure, or a mixture of the two, the E needs to see his results immediately to be able to plan his next study. For this purpose, he needs a rapid, simple, relevant analysis of the data. We define a version of this data-summarizing problem and describe a monitor program to solve it. One major decision is what search method and data file organization to use; We use the n-dimensional array method. We specify dependent variables in two parts: first, a measure code like proportion correct, reaction time, etc., and second, a series of coefficients to specify what items will make up the desired dependent measure. Our program is interactive and makes extensive use of default options. Some suggestions are made for the implementation of monitor programs on other hardware.